The past several decades have seen an alarming rise in the number of infections caused by bacteria resistant to at least one antibiotic. During the same period, there has been an equally alarming decline in the number of new antibiotics receiving approval for clinical use. One reason underlying both trends is that current genomic and other analyses have produced disappointingly few new single drug targets within bacteria, a situation that calls for renewed efforts to develop novel drug target identification strategies such as rational ways to develop combination therapies. We propose one such method here: to deploy genome-scale in silico models of metabolism and other tools from systems biology to identify synthetic lethal and synthetic sick gene pairs within the metabolic networks of pathogenic Enterobacteria. Model predictions would then be tested experimentally by constructing putative pairs in Escherichia coli and Salmonella enterica serovar Typhimurium. Pairs confirmed to be synthetically lethal in both organisms would then be subjected to virtual and high-throughput screening to identify broad-spectrum two-component formulations which inhibit growth or kill multiple members of Enterobacteria. This program would achieve two important goals as a result: to establish systems biology-based metabolic models as one way to uncover - rationally, comprehensively, and in an unbiased manner - targets for combinatorial drug development, and to identify specific pairs of small molecules for possible drug development against an important class of human pathogens.